Emulex Linux Drivers

I have an lp8000 HBA running under the 2.6.7 kernel on the i686 architecture. Built and installed the lpfcdriver-2.6-8.0.9. When I modprobe the lpfc I get the following errors and am unable to see the attached FC devices. Any idea what is wrong?

Aug 18 14:00:31 tng3-1 kernel: lpfc 0000:01:02.0: 0:0321 Unknown IOCB command Data: x0 x3 x0 x0
Aug 18 14:00:31 tng3-1 last message repeated 33 times
Aug 18 14:00:31 tng3-1 kernel: lpfc 0000:01:02.0: 0:1303 Link Up Event x1 received Data: x1 x1 x0 x0
lpfc 0000:01:02.0: 0:0222 Initial FLOGI timeout
lpfc 0000:01:02.0: 0:1303 Link Up Event x5 received Data: x5 x0 x0 x0
Aug 18 14:00:52 tng3-1 kernel: lpfc 0000:01:02.0: 0:0222 Initial FLOGI timeout
Aug 18 14:00:52 tng3-1 kernel: lpfc 0000:01:02.0: 0:1303 Link Up Event x5 received Data: x5 x0 x0 x0
Aug 18 14:01:01 tng3-1 kernel: scsi0 : Emulex LightPulse LP8000 1 Gigabit PCI Fibre Channel Adapter on PCI bus 01 device 10 irq 5
lpfc 0000:01:02.0: 0:0222 Initial FLOGI timeout
Aug 18 14:01:13 tng3-1 kernel: lpfc 0000:01:02.0: 0:0222 Initial FLOGI timeout
lpfc 0000:01:02.0: 0:0127 ELS timeout Data: x4 xfffffe x8a x22
Aug 18 14:01:15 tng3-1 kernel: lpfc 0000:01:02.0: 0:0127 ELS timeout Data: x4 xfffffe x8a x22

============
[root@tng3-1 /]# cat /proc/scsi/lpfc/0
Emulex LightPulse Fibre Channel SCSI driver 8.0.9
Emulex LightPulse LP8000 1 Gigabit PCI Fibre Channel Adapter on PCI bus 01 device 10 irq 5
SerialNum: 0000c9261bf5
Firmware Version: 3.90A7 (D2D3.90A7)
Hdw: 2002506d
VendorId: 0xf80010df
Portname: 10:00:00:00:c9:26:1b:f5 Nodename: 20:00:00:00:c9:26:1b:f5

Link Up - Ready:
PortID 0x0
Point-2-Point
Current speed 1G

This looks like the device on the other end is fairly old and out of date - at least relative to it's support for point-to-point logins and/or it's FC port state machine.

What's happening is that the lp8000 is sending the FLOGI on a point to point link. By the statemachine in the standard - we should receive a response to the FLOGI - and the response should indicate whether there is a fabric present or if there is no fabric, but rather a remote port. The device on the other end of the link is discarding the FLOGI and sending no response. As such, our port state machine does not progress. There does appear to be timers kicking in that resets the link, trying to get this process to continue, but it will have the same results.

There are a couple ways to resolve this:
- Update the firmware on the device on the other end of the link.
- Change the config so it comes up in loop mode rather than pt-2-pt. To do this on the driver, you can set the module parameter "topology=0x04".

I tried modprobing with the lpfc_topology parm and am now getting these errors. Still looks like it is not logging in:

`modprobe lpfc lpfc_topology="0x04"`

lpfc 0000:01:02.0: 0:0321 Unknown IOCB command Data: x0 x3 x0 x0
lpfc 0000:01:02.0: 0:0321 Unknown IOCB command Data: x0 x3 x0 x0
lpfc 0000:01:02.0: 0:0321 Unknown IOCB command Data: x0 x3 x0 x0

lpfc 0000:01:02.0: 0:1303 Link Up Event x1 received Data: x1 x1 x0 x2
Aug 24 13:41:51 tng3-4 last message repeated 143 times
Aug 24 13:41:55 tng3-4 kernel: lpfc 0000:01:02.0: 0:1303 Link Up Event x1 received Data: x1 x1 x0 x2
lpfc 0000:01:02.0: 0:0222 Initial FLOGI timeout
Aug 24 13:42:16 tng3-4 kernel: lpfc 0000:01:02.0: 0:0222 Initial FLOGI timeout
Aug 24 13:42:16 tng3-4 kernel: scsi4 : Emulex LightPulse LP8000 1 Gigabit PCI Fibre Channel Adapter on PCI bus 01 device 10 irq 5
lpfc 0000:01:02.0: 0:0127 ELS timeout Data: x4 xfffffe x8a x21
Aug 24 13:42:18 tng3-4 kernel: lpfc 0000:01:02.0: 0:0127 ELS timeout Data: x4 xfffffe x8a x21

The device at the other end of the link is a Brocade Silkworm 2400 with Fabric OS v2.2.1. Anyone know of publicly available firmware for these things? I tried the Brocade Connect site and obviously don't have their supersecret passkey to access this stuff.

The switch is saying this for the port in question:
port 3: id Online L-Port 1 private, 1 phantom

Another problem I'm having is after I make build && make install the driver and then reboot I'm getting this kernel Oops. Any idea why this is? Thanks in advance for any help you can give me.

Enabling local filesystem quotas: [ OK ]
Enabling swap space: [ OK ]
INIT: Entering runlevel: 3
Entering non-interactive startup
Applying Intel IA32 Microcode update: [ OK ]
Checking for new hardwareOops: 0000 [#1]
Modules linked in: sg lpfc scsi_transport_fc scsi_mod microcode dm_mod ohci_hcd ext3 jbd
CPU: 0
EIP: 0060:[<00000000>] Not tainted
EFLAGS: 00010206 (2.6.7)
EIP is at 0x0
eax: cef46000 ebx: cf144658 ecx: cf144658 edx: cf144658
esi: cef461a0 edi: cef461a0 ebp: cef460a0 esp: cf25beec
ds: 007b es: 007b ss: 0068
Process modprobe (pid: 973, threadinfo=cf25a000 task=cf19f150)
Stack: d08e8fe6 00000002 cef46188 cef46010 cef46000 00000246 cef46000 cff43844
cf25a000 d08fd2c1 ce9512c0 cff43800 d0907844 d0907844 c01b3d68 cff43844
c01fef16 d0907890 d0907890 c01fef38 c031e66c c031e620 c01ff1b1 d090784c
Call Trace:
[<d08e8fe6>] lpfc_sli_hba_down+0x136/0x290 [lpfc]
[<d08fd2c1>] lpfc_pci_remove_one+0x81/0x128 [lpfc]
[<c01b3d68>] pci_device_remove+0x28/0x30
[<c01fef16>] device_release_driver+0x56/0x60
[<c01fef38>] driver_detach+0x18/0x30
[<c01ff1b1>] bus_remove_driver+0x51/0x90
[<c01ff584>] driver_unregister+0x14/0x3a
[<c01b3f7b>] pci_unregister_driver+0xb/0x20
[<d08fd3da>] lpfc_exit+0xa/0x14 [lpfc]
[<c012cabe>] sys_delete_module+0x11e/0x170
[<c014291a>] unmap_vma_list+0x1a/0x30
[<c0142c69>] do_munmap+0x109/0x150
[<c01168b0>] do_page_fault+0x0/0x4cc
[<c0105cad>] sysenter_past_esp+0x52/0x71

Code: Bad EIP value.
<6>inserting floppy driver for 2.6.7
Floppy drive(s): fd0 is 1.44M
FDC 0 is a post-1991 82077
e100: Intel(R) PRO/100 Network Driver, 3.0.18
e100: Copyright(c) 1999-2004 Intel Corporation

I get the same errors on my card as well.
I have an EMC DS-16B (Brocade Silkworm 2800)
FC_Switch:admin> version
Kernel: 5.3.1
Fabric OS: a2.3.0
Made on: Mon Jan 29 18:03:55 PST 2001
Flash: Mon Jan 29 18:05:00 PST 2001
BootProm: Mon Mar 6 20:02:03 PST 2000

Hopefully this is recent enough, if it isnt I have to go through EMC tech support to get a version that is newer.
However, I believemy errors are simply because there is no zoning set up in the switch. I'm learning how to zone it as I type this. :-)

Re danderson - modprobing w/ topology=0x4:
The "Unknown IOCB command Data: x0 x3 x0 x0" concerns me. We've seen this on adapters since we started enabling a reset upon attachment - and it seems to be very prevelant on the lp8000. So far, we've seen this resolved with updating to the latest firmware (see http://www.emulex.com/ts/dds.html). As for the rest of the trace- it's not good that there were so many link up events and that the FLOGI timed out. You shouldn't be having these problems with the Brocade switch. Initially ensure the mode for the Brocade switch port is set for F_Port mode, not G_Port (which tries to autoselect between F & E - as this sometimes causes headaches). Also, the adapter firmware update may aid this.

Re danderson - the kernel oops:
I believes this maps to a problem that is corrected by the 8.0.10 rev of the driver.

Re: Daniel Weigert - same errors:
You shouldn't get FLOGI timeout errors if it's a zoning issue. Any more info as to the "errors" you are seeing.

James,

I have updated the Brocade firmware to v2.6.2a. This did not change things in my setup. Now trying to update the firmware on the lp8000 to 3.92a2. The firmware update package requires lputil, which it says is included in the driver. It does not appear to be included with the driver. Do you know how to get this utility?

Thanks again.
Derek

Ugh... the utilities for the 8.x driver are lagging - waiting for the base driver to settle before they are ported.

What this unfortunately means is that - in the short term - you need to either boot an older driver that does have utilites, or build a dos floppy with the firmware and the f/w update utility - lp6dutil.

So here's the recommendations:
- Follow the firmware links: obtain the lastest firmware image. I believe the image contains the DOS lp6dutil.exe program. Build a bootable DOS floppy, and copy the lp6dutil.exe file, and the *.awc files. Boot the floppy, and use the utility to download the firmware. Additional instructions should be found on the web site.

- Alternatively, if you can boot a 2.4 kernel, you can download the 2.01g driver kit, which includes the necessary utilities. Use the utilities, along with the latest firmware image to update the adapter.

Sorry this is cumbersome right now. This will not be a clean process until kernel acceptance is closer and Emulex's on GR process takes place to create a final 8.x kit to supply the additional utilities.

James,

I backed up to a 2.4 kernel and used lputil to update the HBA firmware, then went back to the 2.6 kernel with the v2.6-8.0.12 driver. Everything is now happy.

Thanks for your help.

I had this same set of errors.

A PtoP connection from linux 2.6 to a brocade switch running 2.6.1c was dumping messages like this:
lpfc 0000:01:02.0: 0:0321 Unknown IOCB command Data: x0 x3 x0 x0
tng3-1 last message repeated 33 times

Upgrading the firmware on the LP8000 to 3.92a2 fixed the problem and the sun has started shining. Luns on the zoned devices are now visable.